Machine Learning (ML) will be key for 5G operational and maintenance solutions. To detect and predict issues with the Network a ML solution will be required.
Machine Learning (ML) predictive solutions will use inputs from the
Mobile Network (4G, 5G, etc..), Operational Support System (OSS), Business Support System (BSS), IT Databases, Device Databases, etc.. the following generalised workflow will be used:-
Pre-processing: Before the data can be utilised by a Machine Learning application, i.e. the learning
algorithm, the data (big / fresh) will need to be "shaped" and normalised.
Pre-processing is a mandatory step for any Machine Learning based solution.
Learning: By comparing different ML algorithms a model can be selected. Via data sets models can be trained and a comparison can be made so that the final optimised model can be used.
Evaluation: Using different training data sets the performance and errors can be estimated.
Prediction: Examining the data result and fine tuning the model.